Optimizing with Genetic

Algorithms
by
Benjamin J. Lynch
T G
C A
A C Feb 23, 2006
G T
T G
T A
G C
C T
 Outline
• What are genetic algorithms?
– Biological origins
– Shortcomings of Newton-type optimizers
• How do we apply genetic algorithms?
– Options to include
• Encoding
• Selection
• Recombination
• Mutation
– Strategies
• What programs can we use?
– How do we parallelize?
• MPI, fork/wait
2
 What are genetic algorithms?
(GAs)
• A class of stochastic search strategies
modeled after evolutionary mechanisms

• a popular strategy to optimize non-linear

systems with a large number of variables

3
 What are genetic algorithms?
(GAs)
• A major difference between natural GAs
and our GAs is that we do not need to
follow the same laws observed in nature.
– Although modeled after natural processes, we
can design our own encoding of information,
our own mutations, and our own selection
criteria.

4
 Definitions for today

• Parameter – a variable in the system of interest

• Gene – encoded form of a parameter being
optimized
• Chromosome – the complete set of genes
(parameters) which uniquely describe an
individual
• Locus – the position of a piece of data within a
chromosome
• Fitness – A value we are trying to maximize

5
 Why would we use genetic algorithms?
Isn’t there a simple solution we learned in
Calculus?

• Newton-Raphson and it’s many relatives

and variants are based on the use of local
information.
• The function value and the derivatives with
respect to the parameters optimized are
used to take a step in an appropriate
direction towards a local maximum or
minimum.
6
 Newton-Raphson

• Perfect for a
parabola!
H x = !g
– H is the Hessian (2nd
derivative with respect
to all parameters
optimized)
– g is the gradient
– x is the step to
minimize the gradient
x

7
 Where Newton-Raphson Fails 
• A local method will only find local
extrema.
If we start our search here
We’ll end up here

H x = !g

8
How do we use GAs to optimize the
parameters we’re interested in?
• Choose parameters to optimize
• Determine chromosomal representation of
parameters
• Generate initial population of individuals
(chromosomes)
• Evaluate fitness of each individual to
reproduce
• Allow selection rules and random behavior
to select next population
9
 Enter
Genetic Algorithm
Create Initial Population

Evaluate the fitness

of each individual Evaluation

Select the parents

based on fitness Selection

Create new population Recombination

10
 Setting up the problem

• This is the most difficult step!

• Choose the parameters you want to optimize

Setup 11
Determine the Parameters

Airfoil Example:
• Constant wing area
• Variable camber
• Variable chord at root
• Variable chord at tip
• Span (function of chords
and wing area)

Setup 12
Density Functional
Example:
• Choose parameters to
be all the variables in
the gradient-corrected
exchange terms.

+ (
) 1/ 3 2 , cx 2
&
) 3+ 3 ( ax , bx e , dx f
2
&" (r ) 4 / 3 dr
E XGGA = ,$ % ) ) & + f & !
2 * 4# ' dx
!
)) 1 + 6ax sinh ,1 x , &&
* Ax '

Setup 13
Evaluate Your Fitness

• For every problem, there is something you

want to maximize or minimize.
• The standard convention is to maximize a
function with a GA.

Setup 14
 Fitness

• We need to evaluate fitness of each

individual.
• The fitness will be used to bias the next
generation towards better genes.

Setup 15
 Fitness

• We must first define what fitness is! You

must come up with a single metric that will
be used to compare 2 possible solutions
and decide which is better.

Setup 16
 • For an airfoil, this might be a function of drag and lift

Fitness = w1 Lift ( P) ! w2 Drag ( P)

• It may depend on a set of simulations at different

speeds, different angles of attack, etc.
simulations simulations
Fitness = ! w Lift ( P) " ! w Drag ( P)
i i
i i

Setup 17
• For an empirical density functional, the fitness
might be a weighted RMS deviation from
experimental values.
dataset
Calc Exp 2
Fitness = " ! (E i ( P) " E
i )
i

Note: we toss in a “–” because we always

Want to maximize the fitness.

Setup 18
• Check that your problem is well-suited for
optimization with a GA.

• Your fitness function will need to be evaluated

thousands of times. Make sure you have the
resources.

• If a GA is too expensive, you still might be able

to simplify your problem and use a GA to find
regions in the parameter space of interest.

Setup 19
• Determine chromosomal representation of
parameters
• Parameters can be encoded in binary, base-
4 base-10, etc.

encoding 20
• After you decide how to encode the parameters,
you must decide on the domain of your
parameters. This is entirely dependent on your
problem. You will want to allow your parameters
to be anything physically reasonable (if you’re
solving a physical problem)

• Create an initial population with randomized

chromosomes

encoding 21
 Create Initial Population
• Population size is chosen (1-10 individuals/parameter
optimized for most applications)
• Parameters to be optimized are encoded.
– Binary, Base 10
– Let’s say we have 2 parameters with initial values of 32
and 13. In binary and base 10 they would look like:

Chromosome of the
100000001101 individual

3213
encoding 22
 How binary genes translate into
parameters
101011101001010111010000100110

698 349 38 You need to

understand the
system you are
optimizing in
order to determine
38 the proper
P = 10 ( Pmax ! Pmin ) + Pmin parameter range
2 !1

encoding 23
 Create Initial Population

• After we choose a population size and

encoding method, we must choose a maximum
range for each parameter. Ranges for
parameters should be determined based on
what would be physically reasonable (if you’re
interested in solving a physical problem).

encoding 24
 • Generate initial population of individuals
(chromosomes)

• The initial population can be generated by

randomizing the genes for each chromosome of
the initial population

• You can set the parameters for a few individuals

if you want. This might speed up the process.

encoding 25
 Review Steps in a GA

1.Initialization of population
2. Evaluation of fitness
3. Selection
4. Recombination
5. Repeat 2-4

26
 Evaluation

• Assign a fitness value to each individual

based on the parameters derived from its
chromosome

Evaluate Fitness 27
• The fitness function is somehow based on
the genes of the individual and should
reflect how good a given set of parameters
is.
– Lift-drag , low drag airfoil
– Ability of a density functional to better predict
chemical phenomena
– Swimming speed of a robotic fish
– Power output of a chemical laser

Evaluate Fitness 28
• Evaluation of the fitness is the computationally-
intensive portion of a GA optimization
• Each chromosome holds the information that
uniquely describes an individual.
• Each chromosome/(parameters set)/individual can
be evaluated separate from the other individuals.
– GA optimizations are typically described as
embarrassingly parallelizable
• The evaluation of the chromosomes reduces down
to a fitness value for each individual which will
be used in the next step

Evaluate Fitness 29
 Selection
• Allow selection rules and random behavior
to select next population

Selection 30
• The parents must be selected based on their
fitness.

• The individuals with a higher fitness must have a

higher probability of having offspring.

• There are several methods for selection.

Selection 31
 Roulette Wheel
Selection
• Probability of parenthood
is proportional to fitness.
• The wheel is spun until
two parents are selected.
• The two parents create Fit(#1)
one offspring. Fit(#2)
Fit(#3)
• The process is repeated to Fit(#4)
Fit(#5)
create a new population
for the next generation.

Selection 32
• Roulette wheel selection has problems if the
fitness changes by orders of magnitude.
• If two individuals have a much higher
fitness, they could be the parents for
every child in the next generation.
Fit(#1)
Fit(#2)
Fit(#3)
Fit(#4)
Fit(#5)

Selection 33
 Another Reason Not to Use the
Roulette Wheel
• If the fitness value for all
individuals is very close,
the parents will be chosen Fit(#1)
Fit(#2)
with equal probability, Fit(#3)
Fit(#4)
and the function will Fit(#5)
cease to optimize.
• Roulette selection is very
sensitive to the form of
the fitness function and
generally requires
modifications to work at
all.

Selection 34
Rank Selection
Fit(#1)
Fit(#2)
Fit(#3)
• All individuals in Fit(#4)

the population are 1 Fit(#5)

ranked according
to fitness
rank
• Each individual is
assigned a weight Fit(#1)
Fit(#2)
inversely Fit(#3)
Fit(#4)
proportional to the 1 Fit(#5)

rank (or other

(rank )1.7
similar scheme).

Selection 35
 Tournament Selection
• 4 individuals (A,B,C,D) are randomly selected from
the population. Two are eliminated and two
become the parents of a child in the next generation

A
A
Fitness(A) > Fitness(B) B

C
D
Fitness(D) > Fitness(C) D

Selection 36
Tournament Selection
• Selection of parents continues until a new
population is completed.
• Individuals might be the parent to several children,
or no children.

A
A
Fitness(A) > Fitness(B) B

C
D
Fitness(D) > Fitness(C) D

Selection 37
 Similarities Between Tournament
and Rank Selection
Fraction of
• Tournament population Fitness
selection is very
similar to rank
9 Both parents were
above the median
selection in the 16
limit of a large
population when 6 One parent was
above the median
we assign a 16
weight of 1/rank.
1 Neither parent was
16 above the median
Selection 38
 Recombination

• Using the chromosomes of the parents, we

create the chromosome of the child

Recombination 39
 Recombination with crossover points
• We can choose a Parent 1
number of Parent 2
crossover points child

A a A In this case, the parameters

Param. 1 B b B remained intact, and the
(eyes) C c C child inherited the same eyes
D d D as parent1 and the same
E e e nose as parent2.
Param. 2 F f f
(nose) G g g
H h h

Recombination 40
crossover point occurs within a parameter
parent1
parent2
child

A a A
Param. 1 B b B
C c C In this case the child will have
(eyes)
D d D a new nose that is not the same
E e E as parent1 or parent2.
Param. 2 F f F
(nose) G g g
H h h

Recombination 41
representation of parameters becomes important.
parent1
parent2 Not possible if
child we used base 10
encoding
1 1 1
Param. 1 1 0 1 1 1 1
1 1 1 5 0 5
1 0 1
0 1 0
Param. 2 0 1 0 0 1 0
0 1 1 0 5 3
0 1 1

Recombination 42
 Recombination – Uniform Crossover
• Uniform crossover
– No limit to crossover points

A a a
B b B Allows more variation in offspring
C c c and decreases need for random
D d D mutations
E e e
F f f
G g g
H h H

Recombination 43
 • Mutations can parent1
be applied after parent2
recombination child

A a A A
Param. 1 B b B B
C c C C
D d D D A random mutation
E e e e has been applied to
Param. 2 F f f f the child
G g g g
H h h h
Random
44
mutations
 • Creep mutations are a special
type of random mutation. Possible creep
• Creep mutations cause a mutations for
parameter to change by a small param. 1
amount, rather than randomizing
any one element.

1 1 1 1 1
Param. 1 1 1 1 1 1
1 1 1 0 1
0 1 0 1 1
OR
0 1 1 1 1
Param. 2 0 0 0 0 0
0 0 0 0 0
0 0 0 0 0

Random
45
mutations
• The desirable frequency of mutations depends
greatly on the other GA options chosen.

A a A A
Param. 1 B b B B
C c C C
D d D D
E e e e
Param. 2 F f f f
G g g g
H h h H

Random
46
mutations
 Other Operators for Recombination
• Other rearrangements of
information are possible
• Swap entire genes
• Swap locus 0 2
4 4 0 5
2 0 4 9
8 8 2 0
5 5 8 3
9 9 5 0
0 0 9 4
3 3 0 2
3 8
Random
47
mutations
 Elitism
• Elitism refers to the safeguarding of the
chromosome of the most fit individual in a given
generation.
• If elitism is used, only N-1 individuals are produced
by recombining the information from parents. The
last individual is a copy of the most fit individual
from the previous generation.
• This ensures that the best chromosome is never lost
in the optimization process due to random events.

48
 More GA Options
• Separate “islands” with populations that interact
infrequently.
• Use “male” & “female” populations
• “Alpha male” selection
• Use 3 parents for each offspring
• Use 3 sexes
• Recessive genes
• …. and many more. Most of which are only useful
for very specific types of fitness functions.

49
 Micro-GA
• Small population size of 1 individual per
parameter optimized.
• No random mutations.
• When the genetic variance is below a
certain threshold (~5%), the most fit
individual goes on, while the chromosomes
of all other individuals are randomized
• Cycle this process

50
Overview of a micro-GA
Create initial population

Evaluate the fitness

of each individual

Select the parents

based on fitness

Recombine parents to
create new population

Is the variance in genes

< than 5% No
Randomize the chromosomes
for all but the most fit individual
Yes
51
 Micro-GA Pros and Cons
• Tends to be very efficient in terms of total
CPU time.
• Robust algorithm with no need to tinker
with random mutation parameters

• It uses a smaller population, and

therefore has less potential for
parallelization.

52
 Review
Evaluation of the fitness is
the time-consuming
portion
Create Initial Population

Evaluate the fitness

of each individual Evaluation

Select the parents

based on fitness Selection

Create new population Recombination 53

 How do you know if you’ve
found the absolute maximum?
• You don’t
• Even GAs are not a “black-box” optimizer
for any function
• You can gain confidence by running
several optimizations with different starting
parameters, different algorithm options,
and different parameter ranges.

54
Which GA options are good picks for
my system?
• Start with robust algorithms
– Micro-GA
– Binary encoding
– Tournament selection
– Uniform crossover at gene boundaries

• If you are unsatisfied with the progress you can

change a couple options
– Allow crossover everywhere to introduce more variety
between each generation.
• Change to base-10 encoding as well, so it isn’t too random
– Use rank selection with a weight of 1/rank3 to heavily
favor the best individuals. 55
How might one parallelize a GA?
• The GA calculations are minimal. An
optimization might require 1000 generations,
and each generation is dominated by the cost
to evaluate the fitness.
• Standard serial GA programs can handle the
GA routines with ~1 ms of cpu time, while
your fitness routine can parallelize the fitness
evaluations.

56
How might one parallelize a GA?
• Evaluating the fitness of an individuals is
independent of the rest of the population.
– You can easily run your GA on N processors
where N is your population size.

• Each individual can often be run in parallel

as well
– This depends on the program you are using to
evaluate the fitness.

57
 Method to Parallelize a GA
• MPI is always a good choice if you’re
already familiar the language. This option
would also enable GA algorithms with
“islands” on heterogeneous clusters.

• Fork/wait would be very easy

– Can be done in several languages

58
#!/usr/bin/perl
use Parallel::ForkManager; Fork method to
$max_process=4;
parallelize with Perl
$pm = new Parallel::ForkManager($max_process); on a shared-memory
@all_chromosomes=(@ARGV)
machine
# enter main loop, no more than $max_process
# children will run at a time
foreach $chromosome (@all_chromosomes) {
my $pid = $pm->start and next;
# prepare an appropriate input file with
# this set of parameters
&writeinput($chromosome);
# run the program to evaluate the fitness
&evalfitness($chromosome);
$pm->finish;
}

# make sure that all the processes are done

$pm->wait_all_children;

#parse output and return to our GA

&parse_output;

59
 GA Programs Available

• David Carrol’s GA Driver

– http://cuaerospace.com/carroll/ga.html
• GAUL
– http://gaul.sourceforge.net/
• GALOPPS (already parallelized)
– http://garage.cps.msu.edu/software/galopps/index.html
• Simple Generalized GA
– http://www.skamphausen.de/software/AI/ga.html

60
Other GA Resources Available

• Me
– blynch@msi.umn.edu
– 612-624-4122

61
Questions?

blynch@msi.umn.edu 4-4122
help@msi.umn.edu 6-0802

62